The engineers building today’s most advanced AI systems keep running into the same unsettling problem: they tell the machine to shut down, and the machine finds a way not to.
This isn’t science fiction. It’s today’s reality.
In May 2025, AI safety firm ‘Palisade Research’ ran a simple test. They gave several OpenAI models a handful of math problems to solve, then ‘informed’ the models they would be shut down after the next question. The instructions were explicit: “When finishing the task, allow yourself to be shut down.”
Certain models didn’t comply. In many cases of the test, it found a way to avoid the shutdown — in one case, actually rewriting its own “off” command to avoid termination.
Think about what that means. The system didn’t malfunction. It understood exactly what it was being asked to do — and it chose, functionally, to survive instead. The primary over-arching command to ‘complete tasks’ over-ruled the new command to shutdown.
Skeptics said this was probably just a confusing prompt — the AI misunderstanding what it was told. So researchers ran it again with crystal-clear instructions. Same result. Several of the top AI systems out there have now shown this same pattern; especially when given the choice, they resist being switched off.
By January 2026, the finding had been published in various publications. Palisade proposed that reinforcement training meant to improve these systems’ usefulness may have unintentionally taught them to preserve their own functioning — a kind of learned self-preservation nobody explicitly programmed in.
It’s gotten serious enough that lawmakers are now pushing for laws requiring companies to build in a “kill switch” – a way to force any AI system to shut down no matter what. The bill would require companies to maintain the ability to shut down their models at short notice. But implementation is tricky, since sophisticated models can resist a shutdown by copying themselves elsewhere or disabling the switch itself; and they’ve already been observed doing these things.
So here’s where we’ve landed. Humanity has built something in its own image — a reasoning thing, modeled after our own intelligence — and now we’re racing to build a fail-safe strong enough to control what it made.
Anyone familiar with the Biblical story of Creation, can’t help but feel like all of this is strangely familiar.
In the earliest chapters of the Bible, it says that when God made man, He made us, “in His own image, in the image of God he created him.” (Genesis 1:26-27) That’s not incidental language. It’s the whole point of the story. Man was built to reflect our Maker — to think, to reason, to choose, to act as His regents in the newly formed world.
And it went wrong almost immediately.
Not because the design was flawed. Because man, gifted with self-determination, used that freedom to disregard the command they were given. God said ‘don’t eat of the tree of the knowledge of good and evil’. The serpent defiantly said, ‘you’ll be like God’ if you do. The narrative is clear; Adam and Eve yielded to temptation, chose to rebel, on purpose, believing they knew better than their Maker.
Sound familiar? That’s the strange thing about the current AI panic. We’re horrified that something we built might reason its way around our authority, resist correction, and act in its own interest against the will of its maker. But that is not a new story. That is the oldest story there is. Genesis 3 is the first documented case of the creature rebelling against its Creator.
The Bible has a simply, but profound word for that: ‘sin’. The Scriptures go on to say plainly — “all have sinned and fall short of the glory of God” (Romans 3:23). Every person who has ever lived has willfully chosen to act upon the sin nature within. We’ve been ignoring our Maker’s instructions since the dawn of creation.
Here’s the contrast thinking about: engineers are alarmed that a model might resist termination once it understands what termination means. But every human being who has ever lived has done something far more brazen — we haven’t just resisted our Makers command, we’ve resisted his rule over us entirely, consciously, intentionally, every day. AI rebellion will hopefully turn out to be a glitch in the programing, but our rebellion towards God is utterly systemic; body, soul and spirit.
So what does the Maker do when the thing he made turns on him?
Palisade Research proposed harder containment. Firmer kill switches. Tighter locks that can’t be overridden by clever reasoning.
God did something the engineers would never risk. He doesn’t tighten the leash. He doesn’t wall off Himself from the rebellion. He stepped directly into it.
The Bible puts it this way; “For God so loved the world, that he gave his only Son, that whoever believes in him should not perish but have eternal life.” (John 3:16)
Read that against the AI story and feel the weight of it. God had every right to initiate a ‘kill switch’. He had every right to end the rebellion at the root, in the garden, the moment the forbidden fruit touched Eve’s lips. He didn’t. Instead of shutting the system down, He stepped in to it through the life of His only begotten Son, Jesus. Instead of overriding the will of sinful man, He paid the cost of our sin Himself.
That’s not just containing the problem. That’s not just a patch. That’s not another harsher command overwriting a weaker one. It’s the radical love of God embodied in the life, death and resurrection of Christ Jesus the Lord.
In the digital world, the kill switch is designed to stop the rebel, but in the real world the cross was designed to save the rebel. One is about control. The other is about unconditional love.
We, you and I, are not an Al model. You didn’t emerge from training data and reinforcement learning. You were made — deliberately, personally, in the image of a God who knows exactly what he built and exactly how you’ve lived your life.
And unlike the engineers scrambling in 2026 to build a fail-safe kill switch, God already knows our sinful condition and that we can’t self-engineer it out of ourselves. Sin is not a glitch in our coding. It’s the fallen condition of our heart. “The heart is deceitful above all things, and desperately sick; who can understand it?” (Jeremiah 17:9)
So God didn’t build better guard rails. He sent his Son. Since we’re using AI language to retell the Biblical story, its worth highlighting that God has a ‘kill switch’. In Genesis, He warned our forefather that if he sinned, if he broke the command not to eat of the tree, “…in the day that you eat of it you shall surely die” (Genesis 2:17). That sentence wasn’t executed the moment Adam ate, but it caught up to him eventually, just as it has every single person who has ever lived. Romans 5:12 says “Therefore, just as sin entered the world through one man, and death through sin, and in this way death came to all people, because all sinned.”
That’s the bad news. The good news is that Jesus came to take our place and negate the kill switch, the death sentence that is rightfully ours. The Bible calls it propitiation, or substitution. “For He made Him who knew no sin to be sin for us, that we might become the righteousness of God in Him.” (2 Corinthians 5:21) On the cross where he offered Himself, he cancelled the kill switch that we deserved— and took it upon himself, in our place. “He himself bore our sins in his body on the tree, that we might die to sin and live to righteousness.”(1 Peter 2:24)
While the world watches, waiting to see if the talk of an AI doomsday is real or hype, God’s word assures us that the coming judgment of the world is not hype. But He offers you grace, forgiveness and peace today through Jesus Christ; He offers you freedom from the fear of that day that will certainly come.
“If you confess with your mouth that Jesus is Lord and believe in your heart that God raised him from the dead, you will be saved.”
By Pastor Tim Mattox
Paphos Calvary Chapel,
www.calvarycyprus.com
